Skip to main content
Engineering blog

Quickly Deploy, Test, and Manage ML Models as REST Endpoints with MLflow Model Serving on Databricks

Aaron Davidson
Tomas Nykodym
Clemens Mewald
Share this post
MLflow Model Registry now provides turnkey model serving for dashboarding and real-time inference, including code snippets for tests, controls, and automation.

MLflow Model Serving on Databricks provides a turnkey solution to host machine learning (ML) models as REST endpoints that are updated automatically, enabling data teams to own the end-to-end lifecycle of a real-time machine learning model from training to production.

Since its launch, Model Serving has enabled many Databricks customers to seamlessly deliver their ML models as REST endpoints without having to manage additional infrastructure or configure integrations. To simplify Model Serving even more, the MLflow Model Registry now shows the serving status of each model and deep links into the Model Serving page.

To simplify Model Serving even more, the MLflow Model Registry now shows the serving status of each model and deep links into the Model Serving page.

To simplify the consumption of MLflow Models even more, the Model Serving page now provides curl and Python snippets to make requests to the model. Requests can be made either to the latest version at a deployment stage, e.g. model/clemens-windfarm-signature/Production or to a specific version number, e.g. model/clemens-windfarm-signature/2

To simplify the consumption of MLflow Models even more, the Model Serving page now provides curl and Python snippets to make requests to the model.

Databricks customers have utilized Model Serving for several use cases including making model predictions in Dashboards or serving forecasts for finance teams. Freeport McMoRan is serving TensorFlow models to simulate operations for their plants:

“We simulate different scenarios for our plants and operators need to review recommendations in real-time to make decisions, optimizing plant operations and saving cost. Databricks MLflow Model Serving enables us to seamlessly deliver low latency machine learning insights to our operators while maintaining a consolidated view of end to end model lifecycle.”

Model Serving on Databricks is now in public preview and provides cost-effective, one-click deployment of models for real-time inference, tightly integrated with the MLflow Model Registry for ease of management. See our documentation for how to get started [AWS, Azure]. While this service is in preview, we recommend its use for low throughput and non-critical applications.

Try Databricks for free

Related posts

Company blog

MLflow Joins the Linux Foundation to Become the Open Standard for Machine Learning Platforms

Watch Spark + AI Summit Keynotes here At today's Spark + AI Summit 2020, we announced that MLflow is becoming a Linux Foundation...
Industries category icon 1

How to Build a Credit Data Platform on the Databricks Lakehouse

Get started and build a credit data platform for your business by visiting the demo at Databricks Demo Center. Introduction According to the...
Platform blog

Simplify Advertising Analytics Click Prediction with Databricks Unified Analytics Platform

July 19, 2018 by Tony Cruz and Denny Lee in Product
Read Rise of the Data Lakehouse to explore why lakehouses are the data architecture of the future with the father of the data...
See all Data Science and ML posts